Flight Measurements of the Aero-Optical EnvironmentAround a Flat-Windowed Turret

Chris Porter,∗ Stanislav Gordeyev,† Mike Zenk,‡ and Eric Jumper§

University of Notre Dame, Notre Dame, Indiana 46545

DOI: 10.2514/1.J052067

This paper discusses the aero-optical environment for a flat-window hemisphere-on-cylinder turret over a wide

range of viewing angles during flight tests in the Airborne Aero-Optics Laboratory. Aero-optical aberrations around

the turretweremeasuredusing a high-speedShack–Hartmannwave-front sensor providing an extensive aero-optical

mapping. The primary data were acquired atMach 0.5 at an altitude of 15,000 ft, with a subset of the data collected at

Mach 0.4 for verification of scaling relationships. Data were acquired holding the relative position between two

aircrafts constant. Additional data sets were acquired allowing two aircrafts to change their relative positions so that

slewing data could be acquired; this provided statistical data over a large range of viewing angles between looking-

forward to looking-back angles. Results were analyzed, and the aero-optical contribution fromdifferent flow features

over the turretwas identified anddiscussed.Cross-correlation functions and convective speedwere also computed for

some viewing angles and compared with other experiments. The flight-test data were also compared to wind-tunnel

measurements using the identical turret.

Nomenclature

AD = aperture diameterAz = azimuthal angleD = turret diameterEl = elevation angleKGD = Gladstone–Dale constantM = incoming Mach numbern = index of refractionOPD = optical path differenceOPDRMS = spatial rms of the optical path differenceOPL = optical path lengtht = times = integration variablex = position vectorx, y, z = turret coordinatesxA, yA = aperture coordinatesα = viewing angleβ = modified elevation angleρ = densityρo = free-stream densityρSL = sea-level densityλ = wavelength

I. Introduction

I N THE late 1970s and early 1980s, optical turrets were exten-sively studied as the use of lasers on aircraft started to become

feasible. During that period, the Airborne Laser Laboratory (ALL)successfully demonstrated the usefulness of airborne lasers [1]. TheALL carbon-dioxide laser’s long wavelength (10.6 μm) limited therange and irradiance that the system could deliver on target. Withnew, more-powerful lasers came the ability to deliver more laserenergy onto a target and an increased range. However, the shorterwavelength (1–1.5 μm) of these new lasers increased the detrimentaleffects that inhomogeneous refractive mediums [2–4] have on theability of optical systems to focus a laser beam in the far field.With the new potential of these lasers, achieving a maximum field

of regard is a necessity. Hemisphere-on-cylinder turrets offer asimple, mechanically efficient means to project or receive laserradiation to or from a target over a full field of regard; however, theflow around a turret consists of a separated wake region aft of theturret, where pressure and temperature fluctuations result in densityand index-of-refraction fluctuations [5,6]. These fluctuations resultin beam jitter (bore-sight error) and higher-order aberrations thatreduce the peak irradiance of the laser beam in the far field. Therefore,aero-optical effects can severely limit an airborne directed-energysystem’s field of regard. To verify wind-tunnel experiments [5] andcomputational simulations [7–10], aero-optical measurements takenunder realistic flight conditions are needed. Yet, to the authors’knowledge, there is no open-literature flight-test data of the flow overa turret available to perform these comparisons. The Airborne Aero-Optics Laboratory (AAOL) [11] reverses this shortfall by addingflight-test capabilities as a viable and affordable experimental tool.This tool not only advances the scientific exploration of aero-opticeffects but also allows for real-flight studies of various mitigationschemes involving flow control and adaptive optics.The AAOL flight program [11] consists of two Citations flying in

formation approximately 50mapart tominimize atmospheric effects.A slowly diverging continuous laser beam approximately 3 mm indiameter is sent from a chase plane to an airborne laboratory; seeFig. 1. The flight-test airborne laboratory consists of a 1-ft-diam(30.5 cm) turret with a 4 in. (10.16 cm) clear aperture; thewindow canbe either flat or conforming to the spherical figure of the turret (i.e.,conformal). The turret itself presents a mold line that is a hemisphereon a cylindrical base, which, when installed in the aircraft, protrudesout the side of the Citation through its modified escape hatch. Theturret can be extended, so that the cylindrical base protrudes intothe airstream by 10.2 cm (which equals the aperture diameter), orwithdrawn, so that only the hemisphere protrudes into the airstream.For the data presented in this paper, the turret was configured with aflat window and extended out the door so that 10.2 cm of thecylindrical base protruded into the airstream. At 50m, the beam fromthe chase aircraft has diverged to approximately 20 cm so that it

Presented as Paper 2011-3280 at the 42th Plasmadynamics and LasersConference, Honolulu, HI, 27–30 June 2011; received 3 May 2012; revisionreceived 26 December 2012; accepted for publication 4 January 2013;published online 23 April 2013. Copyright © 2013 by Porter, Gordeyev, Zenkand Jumper . Published by the American Institute of Aeronautics andAstronautics, Inc., with permission. Copies of this paper may be made forpersonal or internal use, on condition that the copier pay the $10.00 per-copyfee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,MA 01923; include the code 1533-385X/13 and $10.00 in correspondencewith the CCC.

*National Research Council Associate, Department of AerospaceEngineering, U.S. Air Force Academy, Colorado Springs, CO 80840.Member AIAA.

†Research Associate Professor, Department of Mechanical and AerospaceEngineering, Hessert Laboratory for Aerospace Research. Associate FellowAIAA.

‡Project Manager, Department of Mechanical and Aerospace Engineering,Hessert Laboratory for Aerospace Research. Member AIAA.

§Professor, Department ofMechanical andAerospaceEngineering,HessertLaboratory for Aerospace Research. Fellow AIAA.

1394

AIAA JOURNALVol. 51, No. 6, June 2013

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

overfilled the turret clear aperture by a factor of 2; the diverged beampresents a nominal spherical wave front at the laboratory aircraft-turret pupil, which passes through the aero-optical disturbance and iscaptured into the turret’s beam train. Once the laser and turret systemsare tracking each other, a 2.0 cm stabilized beam emerges from theturret mount onto the optical bench in the laboratory aircraft. Thespherical figure of the beam due to the divergence from the chaseaircraft is removed from the otherwise aero-optically aberrated beam.The “stabilization” of the beam is performed by a closed-loop fast-steering-mirror system that reimages the turret pupil and is able toreduce the beam’s overall jitter to a cutoff frequencyof approximately200 Hz, thus acting as a high-pass jitter filter. The “stabilized beam”

is then split to the various sensors on the optical bench onboard thelaboratory aircraft. Details of the experimental setup will be dis-cussed later in this paper.Given the complexity of both the flowfield and experimental setup

to measure the aero-optic effects around an optical turret, manyfundamental studies have been performed by concentrating on thepredominant flow features surrounding a turret, such as separatedshear layers [12–15] and turbulent boundary layers [16–19].Experimental optical data on flat-windowed turrets is limited [5]; toobtain relevant optical data requires facilities large enough forpertinent Reynolds and Mach number flows with optical access.Structural interference from wind tunnels with these capabilities stilllimits the achievable azimuth and elevation angles where optical datacan be acquired.As such, theAAOLalso provides a unique capabilityfor obtaining not only flight data but also ranges of azimuth andelevation angles not obtainable in tunnels due to tunnel blockage andinterference effects. Although the Mach number limit for theCitations is approximately 0.7, because the flow becomes sonic overthe fully protruding turret at flight Mach numbers above 0.55, theAAOL also provides the opportunity of investigating the transonicregime that would require the use of less-available tunnels withdifficult installation requirements. This paper covers data up to onlyMach 0.5; some results of transonic testing can be found in [20,21].As we have mentioned, the flow around hemisphere-on-cylinder

turrets is complicated; the flow characteristics and their relationshipto aero-optical effects are covered in Sec. II. The experimental setupsfor the flight and tunnel tests are discussed in Sec. III. Section IVpresents the optical results from several flights, verifies previouslyproposed scaling relationships, and presents complementary wind-tunnel results. The conclusions are discussed in Sec. V.The flow around a hemisphere-on-cylinder turret is highly com-

plex and three-dimensional [5,22]. The dynamics of the flowfielddependupon both theReynolds number and theMach number. Still, afew general flow features are apparent for most Reynolds numberand fully subsonicMach number tests (fully subsonic in the sense thatthe flow is subsonic everywhere over the turret). A full description ofthe current understanding of the wake region around the turret basedon different tunnel experiments and computational-fluid-dynamicssimulations is presented in a recent review paper [5] and illustrated inFig. 2.At the front of the turret, a necklace vortex forms near the base,

extends along the sides of the turret, and continues in the downstreamdirection aft of the turret. On the upstream half of the turret, the flowremains attached, and an inviscid approximation provides a fairdescription of that portion of the flowfield; however, there is still an

unsteady component that deviates from the inviscid-flow approxi-mation that is due to unsteadiness introduced on the flow by thenecklace vortex. That flow, which is not swept up into the necklacevortex, although somewhat unsteady due to the presence of thenecklace vortex, is forced to accelerate as it goes around and overthe turret with a favorable pressure gradient on the upstream halfof the turret. At the end of the upstream portion and at the aft portionof the turret, the flow experiences an adverse pressure gradientand separates at some point. The location of the separation point andthe downstream wake characteristics are dependent on the Reynoldsnumber. As the flow separates, creating a shear layer, coherentvortical structures form and eventually roll up into two large “horn”vortices. The low pressure and associated low density inside thesevortical structures result in a spatially and temporally varying index-of-refraction field, thus creating a highly aberrating opticalenvironment.Depending on the elevation and azimuthal angle, the beam

entering or exiting the pupil of the turret propagates through some ofthese different regions of the flowfield; see Fig. 2. The location of thepupil is most commonly described using azimuthAz and elevationElangles; however, from a flow-physics point of view, it is moreconvenient to introduce a different coordinate system to describe thebeam direction. This coordinate system uses the viewing angle α andthemodified elevation angle β to define the direction of the beam; seeFig. 3. The viewing angle α is defined as the angle between the flow-direction and beam-direction vectors. The modified elevation angle

Fig. 1 AAOL: a) two citations flying in formation to measure the aero-optic effects of the flow around a turret, and b) picture of the laser spot on theturret.

Fig. 2 Schematic of the subsonic flow around a turret [5].

Fig. 3 Definition of angles to describe the beam direction and turret-and aperture-based system of coordinates.

PORTER ETAL. 1395

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

β is defined as the angle between the junction plane joining thehemisphere to the cylinder and the plane formed by the flow-directionand the beam-directionvectors; notice in Fig. 3 that the junction planeis parallel to the surface on which the cylindrical base is attached.These angles are related to the azimuth and elevation angles throughthe following relations:

α � cos−1�cos�Az� cos�El�� β � tan−1�tan�El�∕ sin�Az�� (1)

The physical reasoning for this coordinate system is based on theobservation that, for flow around a sphere, the time-averaged flowquantities are a function of the viewing angle α only. The addition ofthe modified elevation angle β accounts for the presence of thecylindrical portion of the turret, the presence of the surface on whichthe turret is mounted, and the location of the horn vortices; in theremainder of this paper, effects associatedwith βwill be referred to asmounting-base or surface-plane effects. The effect of the flat-windowat the turret pupil on the asymmetry of the flowfield will also be seenin the modified elevation angle.It is also convenient to introduce two systems of coordinates: one

for the turret and another for the aperture. For the turret-based systemof coordinates (x, y, z), the x axis is chosen to be parallel with the flowdirection, the z direction is along the turret vertical axis, and the y axisis orthogonal to both the x and z axes; see Fig. 3. The aperture-basedsystemof coordinates (xA, yA; see Fig. 3) has the xA axis parallel to thejunction plane, and the yA axis is orthogonal to it pointing upward.As mentioned previously, the unsteady wake generated by the

flow around the hemisphere-on-cylinder turret results in pressure,temperature, and density fluctuations. Through the Gladstone–Daleconstant KGD, the density ρ and the index-of-refraction n are relatedby

n�x; y; z; t� − 1 � KGDρ�x; y; z; t� (2)

Given a variable refractive index flowfield, an initially collimated,planar wavefront laser beam that propagates through the flowfieldemerges with an aberrated wave front. A wave front is defined as asurface of constant phase; if the mean phase is removed, then thewave front is described by a surface of constant phase error (from themean); that surface is displaced from the mean by a distance aboveand below the line of constant mean phase. It can be shownmathematically that the magnitude of the distance above and belowthe line of constant phase is equal to the magnitude of the surfacedefined by the optical path difference (OPD); the two are, in fact,conjugates of one another. If we assume that the propagationdirection is in the z direction, then the OPD is calculated by firstobtaining the optical path length (OPL), which is calculated as

OPL�x; y; t� �Zs2

s1n�x; y; z� dz (3)

from which the OPD is found by

OPD�x; y; t� � OPL�x; y; t� − hOPL�x; y; t�i (4)

where angled brackets represent the spatially averaged optical pathlength over the aperture, x, y, at a given time, t.

II. Flight Experiments

As described in the introduction, a flight experiment consisted ofa laser source aircraft and the main, data-acquisition laboratoryaircraft. The data-acquisition aircraft (the AAOL itself) had both theprotruding turret and the instrumentation for characterizing the aero-optic effects associatedwith the turret’s azimuth and elevation angles.Once in formation at the proper distance and the desired azimuth andelevation with respect to the turret, the laser source aircraft irradiatedthe turret pupil, causing the beam to pass through the aero-opticalflow and enter (in this case, the turret’s flat-window; Fig. 4). Thebeam reflected off the primary and secondary mirrors, which con-tracted the beam and directed it into the Coude path, where a secondtelescope removed the nominal spherical curvature. The overallcontraction took the beam from the clear aperture of 10.2 cm indiameter to the benchwhere it was now 2.0 cm in diameter. The beamwas then reflected off the fast steering mirror (FSM) to removeoverall jitter and onto the optical bench. The beamwas then split by abeam splitter into two beams; see Fig. 5. One beam was sent to aposition-sensing device (PSD), which measured the residual jitter inthe beam after the FSM. Also, the signal from this PSDwas used in aclosed-loop control FSMsystem,which stabilized the outgoing beamfor frequencies up to 200 Hz, as vibration measurements of theoptical table in flight did not reveal any significant frequency contentabove 100 Hz.The second beam was transmitted toward a high-speed Shack–

Hartmann wave front sensor (WFS) through a set of relay optics toreimage the turret pupil onto the lenslet; see Fig. 5. The Shack–Hartmann sensor used for these tests can be operated at capture ratesup to 100 kHz, with the maximum framing rate depending on thenumber of charge-coupled device (CCD) pixels interrogated, whichin turn affects the number of subapertures that can be used on thesensors lenslet array. The full CCD array would allow for a spatialresolution of 60 × 71 lenslets with a maximum capture rate of 7 kHz;on the other hand, if 32 × 32 lenslets were used, thewave fronts couldbe captured at up to 20 kHz. For the majority of the tests, 32 × 32lenslets framed at 20 kHz were used for the fixed-angle data, and32 × 32 lensles framed at 3 kHz were used for the slew data. Tominimize spatial smearing, the camera’s integration time was set at0.5 μs. For the results presented in this paper, from each of thewave-front datasets, the mean aberration from the time series (referred to assteady-lensing aberration) was removed, as well as the instantaneous(i.e., for every frame) piston and tip/tilt (i.e., the residual jitter notremoved by the FSM) leaving only higher-order aberrations.In addition to the high-speedwave-frontmeasurements, Fig. 5 also

shows the schematic of additionalmeasurementsmade during flights.The instantaneous azimuth and elevation angles of the turret wererecorded simultaneously with aero-optical measurements. Also, aboundary-layer Pitot rake was installed below the turret; see Fig. 4,left picture, fromwhich the total and static pressures of the freestreamwere collected to calculate the freestream Mach number usingisentropic compressible relation. The boundary-layer thickness wasalso measured by the Pitot probe rake to be approximately 50 mm.All wave-front measurements referred to here were collected at an

altitude of 15,000 ft for Mach numbers of either 0.4 or 0.5. Wave-front data for this paper were collected for two flight modes, the firstwith the two aircraft fixed in their relative positions to one another,

Fig. 4 Photographs of the turret mounted on the Citation with a schematic of the turret assembly and shell of the aircraft.

1396 PORTER ETAL.

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

which fixed the azimuth and elevation angles (later called fixed-angledata), or with one aircraft passed by the other, which caused the datato have a time-varying azimuth and elevation angles, later referred toas slewing data. For the fixed angle, the wave-front sensor acquireddata at the 20 kHz frame rate for 0.5 s (10,000 wave fronts).For the slewing data, the frame rate of the camera was reduced to

3 kHz, and the sampling time was increased to 14 s; this allowed therapid collection of wave fronts over a range of viewing angles toobtain reliable wave-front statistics, like time-averaged OPDRMS,etc., at these viewing angles. During this time interval, the lasersource aircraft reduced its speed, allowing the laboratory aircraft toslowly pass the source aircraft so that the azimuth/elevation anglesbetween the two planes slowly and monotonically changed. Anexample of the time record of the azimuth/elevation angles during aslewing maneuver is shown in Fig. 6. The elevation angle during thismaneuver remained nearly constant, while the azimuth angle andthus the viewing angle increased almost monotonically by 15–20 deg. During each slewing maneuver, 42,000 wave fronts wererecorded. For the analysis, each dataset was split into sequentialsubsets of 3,000 wave fronts (1 s subsets). For each subset, the time-averaged wave front and the instantaneous tip/tilt were removed, andthe spatial statistics, such as the time-averaged spatialOPDRMS, werecalculated. For each 1 s subset, the corresponding time-averagedazimuth and elevation angles of the flat-windowed aperture wererecorded. A total of 15 slewing maneuvers were performed duringreported flight tests.Typical flow time scales around a turret at these Reynolds and

Mach numbers were on the order of a few milliseconds, while theangular rate of rotation of the turret during a slewing maneuver wason the order of a degree per second. Therefore, at every moment, the

flow quickly adjusted to the slow-changing aperture location, and theoptical statistics (i.e., time-averaged OPDRMS) were not affected bythe rotation of the turret; thus, the slewing maneuvers allowed for arapid mapping of the aero-optic environment around the turret,yielding a large amount of statistical data over a range of azimuth andelevation angles. To verify that the rotation of the turret during thedata-acquisition process did not influence the optical statistics, Fig. 7shows the convergence of the mean spatialOPDRMS of a fixed-angledataset and the corresponding subset of a slewing maneuver at thesame azimuth/elevation angles. Only a small 1% difference existsbetween the converged mean of the fixed-angle test and subset ofthe slewing maneuver; similar results were observed for other datasets investigated. Therefore, statistically breaking up the slewingdata into small subsets of data is equivalent to acquiring a series oftime-uncorrelated fixed-angle data points. Furthermore, higher-orderstatistics, up to the fourthmoment, were calculated andwere found tobe very similar between fixed-angle and slewing subsets. Therefore,unless mentioned otherwise, no distinction will be made between thefixed-angle and slewing-maneuver data.In addition to the flight tests, aero-optical measurements around

the same turret were performed in the University of Notre Dame’s0.9 × 0.9 m wind tunnel to compare the tunnel data with the flightdata. The wind tunnel is a closed-loop, temperature-controlledwind tunnel, capable of freestream Mach numbers up to 0.67 withno models in the tunnel, and it has a turbulent intensity below0.05% [23]. For the tunnel tests, the freestream Mach number was0.3. The turret assemblywas tilted and placed under the test section insuch away that the turret was protruded normally through the bottomtest section wall by the same amount as in flight. The boundary-layerthickness was 30mm upstream of the turret, which is similar with the

Fig. 5 Optical setup of the beam path and suite of instruments during flight tests.

0 5 10 150

50

100

150

Time (s)

Ang

le (

º)

Azimuth AngleElevation AngleViewing Angle

Fig. 6 Example of the instantaneous azimuth, elevation, and viewingangles of the turret during a slewing maneuver.

0 0.2 0.4 0.6 0.8 10.05

0.06

0.07

0.08

0.09

0.1

0.11

Time (s)

mea

n O

PD

RM

S (

µ m)

Fixed-angle convergenceSlewing convergence

Fig. 7 Convergence of the mean spatial OPDRMS of a fixed-angle testand the 1 s subset of a slewing maneuver corresponding to the sameazimuth and elevation angles.

PORTER ETAL. 1397

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

boundary-layer thickness of 50 mm in flight. The experimental setupand pictures of the turret installed in the test section are presentedin Fig. 8. The slowly diverging laser beam traveled approximately50 m, matching the distance between planes in flight, before it wasforwarded to the turret aperture. The turret was actively tracking thebeam, so that the steering mirror was moved to different positions toachieve a range of elevation angle between 40 and 50 deg and a rangeof azimuthal angles between 90 and 130 deg. After exiting the turretassembly, the beam was contracted and forwarded to the high-speedsensor the same way as it is done in flights; see Fig. 5.

III. Results

A. Flight Data

To gain a better understanding of the aero-optic environmentaround the flat-windowed turret, time-averaged spatial variations ofaero-optical distortions OPDRMS over a large range of azimuth andelevation angles were measured and presented in Fig. 9. In addition,Fig. 9a shows the locations of all the elevation/azimuth points (lightdots) where aero-optical distortions were measured. Figure 9b showsthe same data points as a function of the viewing and modified-elevation angles; Figs. 9c and 9d show the data as a location of theaperture center on the surface of the turret, flow direction is along xaxis. In all these figures, the normalized spatial OPDRMS,

OPDNorm�μm∕m� � OPDRMS

�ρ0∕ρSL�M2D(5)

was calculated, and the data in between points were linearly re-interpolated between the closest neighbors to create a continuoussurface to fill in the appropriate gaps. Note that the units of thenormalized OPDRMS are microns per meters. Analysis of experi-mental error showed that experimental error was approximately

0.2 μm∕m, mostly independent of the elevation/azimuthal angles.For clarity, experimental error bars will not be plotted in relatedfigures.Analyzing the results in Fig. 9, at forward-looking angles, Az <

70 deg or α < 70 deg, the normalized OPDRMS remained around 1,which was comparable with the sensitivity of the wave-front sensor.This indicates that the aero-optical environment at these angles wasnot largely affected by higher-order aero-optic aberrations. However,beyond an azimuth (or viewing) angle of 70 deg, the normalizedOPDRMS began to increase, reaching a localmaximumbetween 2 and2.5 at approximately 90 deg. As shown in Fig. 9b, the normalizedOPDRMS was only a weak function of the modified elevation angle βfor viewing angles below 90 deg. The location of the first peak in thenormalized OPDRMS changed by only a few degrees within a rangeof all modified elevation angles tested; however, in the azimuth/elevation angle space, the azimuth location of the first peak changedby roughly 20 deg over tested elevation angles. This illustrates theusefulness of theviewing-angle space, defined in Fig. 3, and indicatesthat mounting-base (or surface-plane) effects over this range ofangles are minimal. After this first peak in the normalized OPDRMS,as the viewing angle continued to increase, the normalized OPDRMS

began to drop slightly; this phenomenon was attributed to tip/tilt-removal effects and is discussed in detail in [24,25].Figures 10a and 10b examine surface-plane effects in more

detail. These are the same data as those presented in Fig. 9, butthey are plotted for various modified elevation-angle regions. InFig. 10a, it can be seen that, for forward-looking angles, thenormalized OPDRMS is largely unaffected by surface-plane effects(i.e., modified elevation-angle effects), although the location of thefirst peak (for α from 90 to 100 deg) shifted slightly as a function ofthemodified elevation angle. The origin of this peakwas discussed inthe previous section. As the viewing angle continued to increasetoward 120 deg, the beam propagated through a larger portion of the

Fig. 8 Turret assembly installed in the tunnel: pictures (top) and experimental setup (bottom).

1398 PORTER ETAL.

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

wake. Simultaneously, the vortical structures in the wake grew largerin this region of the wake, and the normalized OPDRMS increasedrapidly, more than doubling over a few degrees. The data in Fig. 10arepresent the data for modified elevation angles from 20 to 80 deg.The data from Fig. 10a have been represented by a single curve inFig. 10b, with the error bars showing not the experimental errors butrather the data variation over allmodified elevation angles in Fig. 10a;however, additional data for modified elevation angles greater than80 deg have been added to Fig. 10b. As can be seen in Fig. 10b, abovea viewing angle of 100 deg, the data show a marked departure fromFig. 10a.Figures 10a and 10b, taken together, allow for a closer analysis of

the deviation from the data collapse simply on viewing angle. Aspointed out previously, the location of the first local maximumaround the viewing angle of 90 deg is dependent upon the modifiedelevation angle, implying that the location of the maximum issensitive to surface-plane effects.More importantly, formodified ele-vation angles less than 80 deg, the normalizedOPDRMS approaches avalue of nearly 6 around the viewing angle of 130 deg; however, for amodified elevation angle greater than 80 deg, the normalizedOPDRMS dropped to a value of 2, demonstrating that a “sweet spot”exists in the zenith plane that is clearly not present for modifiedelevation angles just 10 deg away from zenith. Computationalsimulations [26,27] have shown that, in this region near the centerlineof the turret (i.e., in the zenith plane), associated with an azimuthangle of 180 deg, the beam propagates through a relatively quietregion of the wake traveling in between the two horn vortices, asdepicted in Fig. 3. This is consistent with centerlinemeasurements on

a canonical flat-windowed turret [28], alongwith conformal-windowturret tests [5], where normalized levels of aero-optical distortionswere also around a value of 2.Returning to the modified elevation-angle data for β < 80 deg,

it is useful to examine the time-averaged and instantaneous wavefronts; Fig. 11 shows the spatial distribution of the wave-fronttemporal variance for data in the region between 60 < β < 70 deg.Figure 11 shows temporal variance of the normalized OPD,� ��OPDNorm�x; y; t��2�

1∕2, where the overbar indicates the temporal

averaging, at every spatial point across the aperture for viewingangles of 81, 92, 99, and 117 deg along with the time-averagedOPDRMS over the entire aperture. The flow across the aperture is fromleft to right. Notice that the temporal variances of the normalizedOPD across the aperture are not constant. Except for the 81 deg case,the magnitude of the fluctuations in the upstream portion of theaperture is smaller than the magnitude of the fluctuations in thedownstream portion of the aperture. This indicates that, as the flowseparates from the turret at back-looking angles, the aero-opticalstructures grow as they propagate downstream, and so anywave-frontmeasurements based on the frozen flow hypothesis, like the Malleyprobe [28], should be treated with caution.Figure 12 shows a selected instantaneous wavefront corre-

sponding to each of the four selected wavefront temporal variationdata sets shown in Fig. 11, flow goes from left to right. In Fig. 12, thescaling in each graph has been adjusted to aid in viewing aberrationsin the wave front. For example, the aberrations shown in Fig. 12a areonly half as large as the aberrations shown in Figs. 12b and 12d.Notice, in Fig. 12a, that the size of the aberrations, relative to

Fig. 9 Time-averaged OPDRMS as a function of a) El −Az angles, b) α − βangles, c) and d) as the location of the aperture center on the turret.

PORTER ETAL. 1399

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

the aperture, are smaller and less coherent than the structures seenat the backward-looking angles. In the wave fronts at backward-looking angles, the aberrations appeared as long spanwise-coherentstructures resulting from the coherent vortical structures in the shearlayer forming as the flow separates over the front portion of theaperture. Additionally, in Fig. 12b, which corresponds to the firstpeak in the normalizedOPDRMS, the upstream portion of the aperturehad a local separation bubble that formed from the slope-discontinuity of the flat-windowed aperture [24,25]. The evidence ofthis unsteady bubble can also be seen in the spatial distribution plot inFig. 11 for the viewing angle of 91 deg.

B. Mach Number Dependence

For the data presented previously, all of the data were normalizedby a scaling factor of �ρ0∕ρSL�M2D. To verify that, at subsonicMachnumbers, this scaling relationship successfully collapsed the data,flight tests were performed at Mach numbers of 0.4 and 0.5. The datawere acquired at 15,000 ft to assure that the only change in the scalingrelationship was the Mach number. Figure 13 shows two data sets oneach plot: one at Mach 0.4 and the other at Mach 0.5. Both data setswere acquired over an azimuth angle range of 145–160 deg and anelevation angle range of 45–67 deg during their respective slewingmaneuvers, so that surface-plane and flat-window effects wereidentical between the two data sets. The top graph shows the unscaleddata, in which the magnitude of the aberration at Mach 0.5 is largerthan the magnitude of the aberration at Mach 0.4. As shown in thebottom graph, the data successfully collapse onto a single curvewhenscaled by �ρ0∕ρSL�M2D, verifying the proposed scaling relationship.Also in Fig. 13, the temporal variation of OPDRMS, defined as�����������������������������������������������������

��OPDRMS�t� − �OPDRMS�2q

, is presented for different viewing

angles as bars. Note that this is not experimental error but a variationof OPDRMS in time. This temporal variation leads to the time-changing far-field intensity, which might impact free-space, laser-based communication systems [20,29].

C. Tunnel Data

Figure 14 shows the comparison between wind-tunnel tests andflight tests at comparable range of angles. For both the flight andtunnel tests, the Reynolds number based on turret diameter wasapproximately 2 million, which is well above the laminar-to-turbulent transition range ofReD � 300; 000 [5]. Both the tunnel andflight data were normalized by �ρ0∕ρSL�M2D.As shown in Fig. 14, the normalization of the data collapses both

the tunnel and the flight data to similar nominal values. The viewingangle of the first localmaximum, corresponding to the presence of theunsteady separation bubble at the leading edge of the flat aperture,has shifted from approximately 102 deg for the flight data to a largervalue of 108 deg in the tunnel tests. The reason for the shift in the peakis currently being investigated, but it is probably related to theboundary layer separating from the turret earlier during flight than inthe tunnel because of tunnel-blockage effects. In the 0.9 × 0.9 mtunnel, the turret has a blockage of ∼7%, where we here hypothesizethat it has changed the separation by ∼6 deg; however, in testing aturret with an even larger tunnel blockage, it was observed that theblockage has delayed separation up to 130 deg [5]. Given thedifference in the elevation angle and blockage effects, the normalized

Fig. 10 Changes in the normalized OPD as a function of viewing anglefor different elevation angle ranges. Bars denote the variation of dataover all modified elevation angles and not experimental error.

Fig. 11 OPDNorm centered between 60 < β < 70 deg with spatial distributions of the wavefront temporal variance across the aperture for selectedangles. Color scales are different for each subplot.

1400 PORTER ETAL.

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

data from the tunnel tests and from the flight tests match fairly well,and the existing differences are mostly within the error of the mea-surements, as the experimental error was estimated to be about0.2 μm∕m.

D. Structure Size and Correlation of the Data

Both streamwise and spanwise correlation lengths can be obtainedby autocorrelating the wave fronts. Figure 15 shows a normalized-autocorrelation two-dimensional map and the corresponding slicesthrough the correlationmap atΔy∕AD � 0 andΔx∕AD � 0; the datashown are in-flight data at an azimuth angle of 134 deg and an

elevation angle of 65 deg, which corresponds to a viewing angle of107 deg. At this viewing angle, the flow separates at the leading slopediscontinuity of the flat aperture and forms a spatially evolving shearlayer over the aperture [26,28]. Using the definition of the correlationlength as the location of the first minima, the structure size in thiscase was approximately 0.3AD. This result is slightly larger than thecorrelation lengths reported in [28], where the streamwise correlationlength was 0.25AD, although the Mach number and viewing anglewere different in that case; furthermore, one-dimensional wave frontswere measured in [28] using a Malley probe, and so spanwise effectswere not captured. Concentrating on the fixed-angle data between60 < β < 70 deg, the structure size was calculated for the variousviewing angles and results are presented in Fig. 16. The structure size,defined as a location of the first minima in the streamwise direction,decreases as the viewing angle increases until it approximately levelsout around 93 deg. This location corresponds to the angle atwhich thenormalizedOPDRMS reaches the first local maxima at these modifiedelevation angles. The correlation length starts increasing for largeviewing angles above 110 deg, consistent with the fact that theseparated region over the aperture and vortical structures inside of itget larger for large viewing angles.

-0.5 0 0.5-0.5

0

0.5 0.5

x/AD

y/A

D

OPDNorm

OPDNorm OPDNorm

OPDNorm

-2

-1

0

1

2

-0.5 0 0.5-0.5

0

x/AD

y/A

D

-10

-5

0

5

10

-0.5 0 0.5-0.5

0

0.5 0.5

x/AD

y/A

D

-6

-4

-2

0

2

4

-0.5 0 0.5-0.5

0

x/AD

y/A

D-10

-5

0

5

10

UnsteadyBubble

Az = 67 deg, El = 68 deg Az = 96 deg, El = 68 deg

Az = 112 deg, El = 66 deg Az = 142 deg, El = 54 deg

a) b)

c) d)Fig. 12 Instantaneous realizations of wavefronts for a) α � 81 deg, b) α � 92 deg, c) α � 99 deg, and d) α � 117 deg. The color map in each part is

adjusted appropriately.

105 110 115 120 125 130 1350

0.1

0.2

0.3

Viewing Angle , deg

Viewing Angle , deg

OP

DR

MS (

µ m)

Mach 0.5Mach 0.4

105 110 115 120 125 130 135

2

4

6

8

OP

DN

orm

(µ m

/m)

Mach 0.5Mach 0.4

α

αFig. 13 Comparison of optical data forM � 0.4 and 0.5, the unscaleddata (top plot), and the scaled data (bottom plot). Bars indicate levels oftemporal variation of OPDRMS.

Fig. 14 Comparison of flight data and tunnel data for the AAOL turretfor β < 45 deg. Experimental error is about 0.2 μm∕m.

PORTER ETAL. 1401

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

Using the time-resolved wave-front data, the convective velocityof optical structures can be calculated in one of several ways: 1) byspatially cross-correlating time-delayed wave fronts over the entireaperture; 2) by calculating the phase slope as a function of thefrequency, similar to the Malley probe analysis [28,30]; 3) byanalyzing spatially temporary-resolved wave fronts in thefrequency–wavenumber space [31]; or 4) by using other methods.Figure 17 shows the average calculated velocity of the opticalstructures based on the first approach. Depending on the timedifference between the wave fronts being correlated, the opticalstructure velocity varied slightly; however, for all the cases, thestructures were primarily propagating in the streamwise direction

across the aperture at a velocity of approximately 0.95 of thefreestream velocity. Similar values were reported in [28]. Note thatthe local flow speed over the aperture is larger than the freestreamvelocity by approximately a factor of 1.5 due to the fact that theflow naturally accelerates as it passes over the turret; this factor wascalculated using static pressure measurements over conformal-window turrets with the same diameter-to-height ratio [5]. Asmentioned previously in relationship to Fig. 14, the presence oftunnel walls not only adds to the acceleration over a turret but alsoaffects the flow angularity, which provides the possible explanationfor the 6 deg shift in the local peak between 90 and 100 deg in Fig. 14.Taking the 1.5 multiplier of the flow over the top of the turret, theconvective velocity of the abberating structures convecting overthe aperturewas approximately 0.6 of the local flow velocity over theaperture. This value is consistent with the normalized convectivespeed of large-scale structures within shear layers and has beenobserved experimentally at back-looking angles in other tests[24,28].

IV. Conclusions

This paper presents the results of the extensive measurements ofthe complex aero-optical environment around the side-mounted, flat-window turret performed in-flight using the Airborne Aero-OpticsLaboratory (AAOL). Before this set of data, optical-experimentaldata of the aero-optical environment around a flat-windowed turret inopen literature were very sparse, with the largest published data setconsisting of only four data points. Furthermore, all data up to thispoint have been obtained either in a wind tunnel or computationally.The AAOL provides a unique capability to collect large amountsof flight data during relatively short flight sequences. Using ahigh-speed wave-front sensor onboard the AAOL allows acquiringwave fronts at different azimuth/elevation angles with good spatialand temporal resolution. To analyze the data, a different frame ofreference, the viewing and modified-elevation angles, was proposed.In this frame of reference, differences in optical data at a fixedviewing angle can be directly attributed to surface-plane effects andasymmetries in the wake around the turret. Results showed that aero-optical aberrations are negligible for small forward-looking viewingangles less than 70 deg, where aero-optical effects are primarily froma less-aberrating attached boundary layer over the turret. Above70 deg, a small separation bubble appeared at the front edge of theflat window, introducing convecting, small-scale structures overthe aperture and causing aero-optical distortions to increase. Theseparation bubble increaseswith the increased viewing angle, leadingto a subsequent increase in aero-optical distortions. The aero-opticallevels have a local peak around the viewing angle between 90 and100 deg, followed by a small drop in the levels in the range of viewingangles between 100 and 110 deg. This drop in the aero-optical

Fig. 15 Normalized autocorrelation of OPD at Az � 134 deg, El � 65 deg, α � 107 deg; incomingM � 0.5.

Fig. 16 Calculated structure size from auto correlatingwave-front datafor 60 < β < 70 deg.

100 150 200 250 300 3500

0.5

1

1.5

∆t between correlated wavefronts (µs)

V C/V

∞

streamwise-componentspanwise-componentmagnitude

Az=134 deg, El = 65 deg

Fig. 17 Calculated velocity by correlating wavefronts using differenttime delays between wavefronts at a flight speed of Mach 0.5.

1402 PORTER ETAL.

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067

distortions, also observed in other experiments, was contributed to thefact that the instantaneous tip-tilt was removed from aero-optical data.Above 110 deg, aero-optical aberrations increase sharply with theviewing angle as the laser beam propagates through a stronger regionof the turbulent wake, dominated by unsteady horn vortices. Theseincreased aero-optical levels were not observed for the region aft of theturret near the center plane with the azimuth angle around 180 deg,where the levels stayed approximately constant or even decreased, asthe flow in this region between horn vortices is relatively quiet.Data sets in flight at different Mach numbers showed that the

scaling relationship normalizing low subsonic optical data by�ρ0∕ρSL�M2D successfully captured the Mach number dependence.In addition to time-averaged levels of aero-optical distortions, cross-correlation maps and mean convective speeds were calculated atselected angles; resultswere found to agree favorablywith previouslyreported measurements. Finally, flight results were compared towind-tunnel data results and differences were discussed.

Acknowledgments

The authors would like to thank the Northern Air personnel,especially Kevin Tessmer for providing technical assistance duringflight test. We would also like to thank Matthew Krizo from the AirForce Institute of Technology (AFIT) for providing assistance on thelaser tracking system.These efforts were sponsored by the High Energy Laser Division

of the Joint Technology Office and supported by the U.S. Air ForceOffice of Scientific Research, grant FA9550-07-1-0574.

References

[1] Duffner, R., The Adaptive Optics Revolution: A History, Univ. of NewMexico Press, Albuquerque, NM, 2009, pp. 202–235.

[2] Ross, T. S., “Limitations and Applicability of the MaréchalApproximation,”AppliedOptics, Vol. 48, No. 10, 2009, pp. 1812–1818.doi:10.1364/AO.48.001812

[3] Butts, R., and Weaver, L. D., “Airborne Laser Experiment (ABLEX):Theory and Simulations,” Proceedings of SPIE 2120, Laser Beam

Propagation and Control, Vol. 10, Los Angeles, CA, June 1994.[4] Weaver, L.D., andButts, R., “ABLEX:HighAltitudeLaser Propagation

Experiment,” Proceedings of SPIE 2120, Laser Beam Propagation and

Control, Vol. 30, Los Angeles, CA, June 1994.[5] Gordeyev, S., and Jumper, E. J., “Fluid Dynamics and Aero-Optics

of Turrets,” Progress in Aerospace Sciences, Vol. 46, No. 8, 2010,pp. 388–400.doi:10.1016/j.paerosci.2010.06.001

[6] Wang, M., Mani, A., and Gordeyev, S., “Physics and Computation ofAero-Optics,” Annual Review of Fluid Mechanics, Vol. 44, Jan. 2012,pp. 299–321.doi:10.1146/annurev-fluid-120710-101152

[7] Ladd, J., Mani, M., and Bower, W., “Validation of Aerodynamic andOptical Computations for the Flow About a Cylindrical/HemisphericalTurret,” 27th AIAA Applied Aerodynamics Conference, AIAA Paper2009-4118, June 2009.

[8] Pond, J. E., and Sutton, G., “Aero-Optic Performance of an AircraftForward-Facing Optical Turret,” Journal of Aircraft, Vol. 43, No. 3,2006, pp. 600–607.doi:10.2514/1.18601

[9] Morgan, P. E., and Visbal, M. R., “Large Eddy Simulation of FlowOvera Flat-Window Cylindrical Turret with Passive Flow Control,” 48th

AIAA Aerospace Sciences Meeting, AIAA Paper 2010-916, Jan. 2010.[10] White, M. D., Morgan, P. E., and Visbal, M. R., “High Fidelity Aero-

Optical Analysis,” 48th AIAA Aerospace Sciences Meeting, AIAAPaper 2010-433, Jan. 2010.

[11] Jumper, E., Zenk, M., Gordeyev, S., and Cavalieri, D., “The AirborneAero-Optics Laboratory, AAOL,” Proceedings of SPIE, Vol. 8395,Paper :8395-6, June 2012.

[12] Fitzgerald, E. J., and Jumper, E. J., “The Optical Distortion Mechanismin a Nearly Incompressible Free Shear Layer,” Journal of Fluid

Mechanics, Vol. 512, Aug. 2004, pp. 153–189.doi:10.1017/S0022112004009553

[13] Rennie, R. M., Duffin, D. A., and Jumper, E. J., “Characterization andAero-Optic Correction of a Forced Two-Dimensional Weakly Compress-

ible Shear Layer, AIAA Journal, Vol. 46, No. 11, 2008, pp. 2787–2795.doi:10.2514/1.35290

[14] Zubair, F. R., Freeman, A. P., Piatrovich, S., Shockro, J., Ibrahim, Y. N.,and Catrakis, H. J., “Large Scale Turbulence Suppression Controlfor Direct Reduction of Aero-Optical Aberrations,” 38th AIAA

Plasmadynamics and Lasers Conference, AIAA Paper 2007-4008,June 2007.

[15] Catrakis, H. J., and Aguirre, R. C., “New Interfacial Fluid ThicknessApproach in Aero-Optics with Applications to Compressible Turbu-lence,” AIAA Journal, Vol. 42, No. 10, 2004, pp. 1973–1981.doi:10.2514/1.547

[16] Cress, J. A., Gordeyev, S., Post, M. L., and Jumper, E. J., “Aero-OpticalMeasurements in a Turbulent, Subsonic Boundary Layer at DifferentElevation Angles,” 39th Plasmadynamics and Lasers Conference,AIAA Paper 2008-4214, June 2008.

[17] Wyckham, C. M., and Smits, A. J., “Aero-Optic Distortion in Transonicand Hypersonic Turbulent Boundary Layers,” AIAA Journal, Vol. 47,No. 9, 2009, pp. 2158–2168.doi:10.2514/1.41453

[18] Wang, K., and Wang, M., “Aero-Optics of Subsonic TurbulentBoundary Layers,” Journal of Fluid Mechanics, Vol. 696, April 2012,pp. 122–151.doi:10.1017/jfm.2012.11

[19] Gordeyev, S., Jumper, E. J., andHayden, T., “Aero-Optics of SupersonicBoundary Layers,” AIAA Journal, Vol. 50, No. 3, 2012, pp. 682–690.doi:10.2514/1.J051266

[20] DeLucca, N., Gordeyev, S., and Jumper, E. J., “In-flight Aero-Optics ofTurrets,” Journal of Optical Engineering, Vol. 52, No. 7, 2013,p. 071405.doi:10.1117/1.OE.52.7.071405

[21] De Lucca, N., Gordeyev, S., and Jumper, E., “Comparison of Aero-Optical Measurements from the Flight Test of Full and HemisphericalTurrets on the Airborne Aero-Optics Laboratory,” 43rd AIAA

Plasmadynamics and Lasers Conference, AIAA Paper 2012-2985,June 2012.

[22] Wallace, R. D., Shea, P. R., Glauser, M. N., Vaithianathan, T., andCarlson, H. A., “Feedback Flow Control for a Pitching Turret (Part 2),”48th AIAA Aerospace Sciences Meeting, AIAA Paper 2010-361,Jan. 2010.

[23] Cress, J. A., “Optical Distortions Caused by Coherent Structures in aSubsonic Compressible, Turbulent Boundary Layer,” Ph.D Thesis,Univ. of Notre Dame, Notre Dame, IN, 2010.

[24] Gordeyev, S., Cress, J. A., Jumper, E., and Cain, A. B., “Aero-OpticalEnvironment Around a Cylindrical Turret with a Flat Window,” AIAAJournal, Vol. 49, No. 2, 2011, pp. 308–315.doi:10.2514/1.J050476

[25] DeLucca,N.,Gordeyev, S., and Jumper, E., “The Study ofAero-Opticaland Mechanical Jitter for Flat Window Turrets,” 50th AIAA Aerospace

Sciences Meeting, AIAA Paper 2012-0623, Jan. 2012.[26] Morgan, P., and Visbal, M., “Numerical Simulation Investigating

Control of Flow over a Flat Window Turret,” 49th AIAA Aerospace

Sciences Meeting, AIAA Paper 2011-468, Jan. 2011.[27] Morgan, P., and Visbal, M., “Simulation of Aero-Optics for Flow over a

Flat-Window Hemispherical Turret,” 41st AIAA Fluid Dynamics

Conference and Exhibit, AIAA Paper 2011-3264, June 2011.[28] Gordeyev, S., Hayden, T., and Jumper, E., “Aero-Optical and Flow

Measurements over a Flat-Windowed Turret,” AIAA Journal, Vol. 45,No. 2, 2007, pp. 347–357.doi:10.2514/1.24468

[29] Gordeyev, S., Cress, J., and Jumper, E., “Far-Field Laser IntensityDrop-Outs Caused by Turbulent Boundary Layers,” Journal of Directed

Energy (to be published).[30] Abado, S., Gordeyev, S., and Jumper, E., “Approach for Two-

Dimensional Velocity Mapping,” Journal of Optical Engineering,Vol. 52, No. 7, 2013, p. 071402.doi:10.1117/1.OE.52.7.071402

[31] Smith, A. E., Gordeyev, S., and Jumper, E., “Recent Measurements ofAero-Optical Effects Caused by SubsonicBoundary Layers,” Journal ofOptical Engineering, Vol. 52, No. 7, p. 071404, 2013.doi:10.1117/1.OE.52.7.071404

L. CattafestaAssociate Editor

PORTER ETAL. 1403

Dow

nloa

ded

by U

NIV

ER

SIT

Y O

F N

OT

RE

DA

ME

on

June

7, 2

013

| http

://ar

c.ai

aa.o

rg |

DO

I: 1

0.25

14/1

.J05

2067